lov-clone-rebrand

Pass

Audited by Gen Agent Trust Hub on Sep 7, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external project templates and forks which could contain malicious instructions designed to influence the agent.
  • Ingestion point: Reads template project files, configuration, and documentation during the rebranding process (Workflow Steps 1-3).
  • Boundary markers: Absent. The skill does not define specific delimiters or warnings to separate ingested content from system instructions.
  • Capability inventory: File system access, Git repository operations, and shell command execution.
  • Sanitization: Absent. While it filters sensitive keys in its profile storage script, it does not explicitly sanitize text ingested from projects before use.
  • [DYNAMIC_EXECUTION]: The skill triggers the execution of deployment and publishing workflows defined within the project it is rebranding.
  • Evidence: Workflow Step 5 specifies using the "current project's publishing process" for deployment tasks.
  • Risk: This pattern involves executing local scripts or commands found within untrusted project templates.
  • [COMMAND_EXECUTION]: The skill uses local shell commands to perform file system operations and Git tasks.
  • Evidence: The execution boundary in SKILL.md explicitly mentions utilizing CLI tools provided by the host environment.
  • [DATA_EXFILTRATION]: The skill is capable of reading local project content and transmitting it to remote Git repositories.
  • Evidence: Workflow Step 4 describes creating and pushing to new GitHub repositories.
  • Note: The skill includes logic in Workflow Step 3 to ensure credential values are not copied during the rebranding process.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 7, 2026, 09:46 AM
Security Audit — agent-trust-hub — lov-clone-rebrand